https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67070
--- Comment #6 from Jason Merrill <jason at gcc dot gnu.org> --- Created attachment 36107 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=36107&action=edit patch to satisfy using pre-normalized constraints When I raised this on the core reflector, Richard Smith said that he thought satisfaction used the pre-normalization constraints. I think that's not what the TS currently says, but making that change fixes this issue and seems reasonable to me. What do you think, Andrew?
